Can Economics-based Resource Allocation Prove Effective in a Computation Marketplace?

Abstract

Several companies offer computation on demand for a fee. More companies are expected to enter this business over the next decade, leading to a marketplace for computation resources. Resources will be allocated through economic mechanisms that establish the relative values of providers and customers. Society at large should benefit from discoveries obtained through the vast computing power that will become available. Given such a computation marketplace, can economics-based resource allocation provide benefits for providers, customers and society? To investigate this question, we simulate a Grid economy where individual providers and customers pursue their own ends and we measure resulting effects on system welfare. In our experiments, customers attempt to maximize their individual utilities, while providers pursue strategies chosen from three classes: information-free, utilization-based and economicsbased. We find that, during periods of excess demand, economics-based strategies yield overall resource allocation that benefits system welfare. Further, economics-based strategies respond well to sudden overloads caused by temporary provider failures. During periods of moderate demand, we find that economics-based strategies provide ample system welfare, comparable with that of utilization-based strategies. We also identify and discuss key factors that arise when using economic mechanisms to allocate resources in a computation marketplace.

Improving efficiency of markov chain analysis of complex distributed systems

“…Now, using the perturbed matrices created by the appropriate perturbation algorithm, the resulting Markov chain was utilized to determine the proportion of tasks completed by this new system. Additionally, the perturbed matrices were used to find the theoretical approximation to the proportion of tasks completed using Equation 6. The value for proportion of tasks completed for the Markov model and the theoretical model were then recorded together with a summary value for the entry (r, c ↓ ) (or (r, c ↑ ), depending on the perturbation method used) from the perturbed TPMs.…”

“…The Markov chain model is derived from a previous large-scale grid computing system model [1,6] that simulates the progress of a large number of computing tasks from the time they are submitted to the grid for execution by an end user to the time they either complete or fail. Figure 3.1 shows this Markov model as a state diagram for a single task.…”

“…Studies of alternative economic strategies [4][5][6] and failure scenarios [1] have shown that small variations in key system variables can lead to large differences in overall system performance. While large-scale simulations are more practical than operational testbeds, computational expense can increase dramatically with model size, a critical factor for studying large-scale systems such as the Internet.…”

“…(1,1) 0.680000 (1,2) 0.020000 (1,4) 0.020000 (1,5) 0.020000 (1,6) 0.040000 (1,7) 0.040000 (1,9) 0.020000 (1,11) 0.020000 (1,13) 0.040000 (1,14) 0.040000 (1,15) 0.020000 (1,17) 0.020000 (1,18) 0.020000 (2,24) 0.040000 (2,25) 0.020000 (2,28) 0.020000 (2,29) 0.020000 (2,31) 0.020000 (2,34) 0.020000 (2,35) 0.020000 (2,36) 0.020000 (2,38) 0.020000 (2,41) 0.020000 (2,42) 0.020000 (2,43) 0.020000 (2,44) 0.020000 (2,…”

“…1.000000 (6,50) 1.000000 (7,50) 1.000000 (8,50) 1.000000 (9,50) 1.000000 (10,50) 1.000000 (11,50) (25,1) 0.360000 (25,3) 0.040000 (25,6) 0.040000 (25,7 ) 0.040000 (25,8) 0.040000 (25,9) 0.020000 (25,11) 0.040000 (25,13) 0.040000 (25,14) 0.020000 (25,15) (26,1) 0.640000 (26,2) 0.020000 (26,4) 0.020000 (26,5) 0.040000 (26,7) 0.020000 (26,10) 0.040000 (26,11) …”